[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bmwg] RE: Methodology for benchmarking fast failover time with local protection Part 2



 Scott,

I read your draft draft-poretsky-mpls-protection-meth-04.txt and realized that we could expand on the FRR benchmarking of the failover times with all possible combination of topologies and additional test cases.

So Jay and I came up with a write-up that could be merged to the existing work of yours. Please take a look at this and let us know what you and the co-authors of draft-poretsky-mpls-protection-meth-04.txt think ? 

I'm sending my draft  fragmented into 2 text files and attached in 2 different emails to stay within the limit of  25K mail size. 

Thanks,

Samir

 




3 FRR Scalability table

The following lists out recommended scaling limits of IGP/VPNv4 Prefixes, LSP Tunnels 
and VC entries. Depending upon the topology the appropriate scaling limits can be 
benchmarked.


3.1  FRR IGP Table

      Test ID No of Headend     IGP Prefixes
              TE LSPs

      3.1.1   1                  100
      3.1.2   1                  500
      3.1.3   1                 1000
      3.1.4   1                 2000
      3.1.5   1                 5000
      3.1.6   2(Load Balance)    100
      3.1.7   2(Load Balance)    500
      3.1.8   2(Load Balance)   1000
      3.1.9   2(Load Balance)   2000
      3.1.10  2(Load Balance)   5000
      3.1.11  100                100
      3.1.12  500                500
      3.1.13  1000		1000
      3.1.14  2000		2000


3.2   FRR VPNv4 Table

      Test ID No of Headend     VPNv4 Prefixes
              TE LSPs

      3.2.1   1                  100
      3.2.2   1                  500
      3.2.3   1                 1000
      3.2.4   1                 2000
      3.2.5   1                 5000
      3.2.6   1                10000
      3.2.7   1                20000
      3.2.8   1                  Max
      3.2.9   2(Load Balance)    100
      3.2.10  2(Load Balance)    500
      3.2.11  2(Load Balance)   1000
      3.2.12  2(Load Balance)   2000
      3.2.13  2(Load Balance)   5000
      3.2.14  2(Load Balance)  10000
      3.2.15  2(Load Balance)  20000
      3.2.16  2(Load Balance)    Max

3.3   FRR Mid-point LSP Table

      Test ID No of Mid-point     
              TE LSps

       3.3.1   100
       3.3.2   500
       3.3.3  1000
       3.3.4  2000
       3.3.5   Max 

3.4   FRR VC Table

Test ID No of Headend     VC entries
              TE LSPs

       3.4.1   1           100
       3.4.2   1           500
       3.4.3   1          1000
       3.4.4   1          2000
       3.5.5   1           Max
       3.5.6   100         100
       3.5.7   500         500
       3.5.8  1000        1000
       3.5.9  2000        2000


4. Test Procedures

The following highlights the various test procedures for benchmarking FRR Failover.


   4.1 Head-end as PLR
   4.1.1 Link Protection (Nhop)

   Objective
     To benchmark the MPLS failover time due to a link failure event experienced by
     egress of PLR(DUT). 

   Test Setup
     Use the topologies in Figure 1 through 4.  The DUT is the LSP ingress.
     The DUT will have 3 interface connected to the traffic generator
     The DUT will also have 2 interfaces connected to the test device
     
   Test Configuration
     1. Configure a primary and the backup for each of the link protection cases
        as shown in Figures 1 through 4.
     2. Provision the tailend to send implicit null label for the primary 
        and backup tunnel
     3. Advertise prefixes (as per FRR Scalability table describe in section 2) by the 
        tail end. 
        
   Procedure 
     1. Configure primary and backup tunnel 
     2. Verify Primary and Backup tunnels are up and that Primary is protected 
     3. Verify Fast reroute protection
     4. Configure 3 traffic stream using any traffic generator. 3 traffic streams should be 
        configure to send traffic to 1st , Mid and Last prefixes from fast reroute database, 
        Make sure do not configure traffic stream in round robin fashion. we need 3 separate 
        stream send to 1st mid and last prefixes 
     5. Start traffic as specified in section "Traffic Generation" 
     5. Shutdown Protected interface from remote side (or any other failure trigger) 
     6. Verify Primary tunnel and prefixes gets mapped to backup tunnels   
     7. Stop traffic stream and measure the traffic loss.
     8. Measure failover time . (Numbers of packet drop/rate per second * 1000)
     9. Start traffic stream again to verify re-optimization when protected interface 
        comes up. Traffic loss should be 0 due to make before break or reversion
     10.Enable protected interface that was shut
     11.Verify headend signals new LSP and Protection should be in place again

   4.1.2 Node Protection (NNhop)
   Objective
     To benchmark the MPLS failover time due to a failure of node downstream to
     egress.
 
   Test Setup
     Use the topologies in Figures 5 through 8.  The DUT is the LSP ingress.
     The DUT will have 3 interface connected to the traffic generator
     The DUT will also have 2 interfaces connected to the test device
     
   Test Configuration
     1. Configure a primary and the backup for each of the Node protection cases
        as shown in Figures 5 through 8
     2. Provision the tailend to send implicit null label for the primary 
        and backup tunnel
     3. Advertise prefixes (as per FRR Scalability table describe in section 2) by the 
        tail end. 
        
   Procedure 
     1. Configure primary and backup tunnel 
     2. Verify Primary and Backup tunnels are up and that Primary is protected 
     3. Verify Fast reroute protection
     4. Configure 3 traffic stream using any traffic generator. 3 traffic streams should be 
        configure to send traffic to 1st , Mid and Last prefixes from fast reroute database, 
        Make sure do not configure traffic stream in round robin fashion. we need 3 separate 
        stream send to 1st mid and last prefixes
     4. Start traffic as specified in section "Traffic Generation" 
     5. Reboot or turn off power of the node downstream to the PLR/Ingress (or any other 
        failure trigger)
     6. Verify Primary tunnel and prefixes gets mapped to backup tunnels   
     7. Stop traffic stream and measure the traffic loss.
     8. Measure failover time . (Numbers of packet drop/rate per second * 1000)
     9. Start traffic stream again to verify re-optimization when protected node 
        comes up. Traffic loss should be 0 due to make before break or reversion
     10.Bring the node that was brought down up
     11.Verify headend signals new LSP and Protection should be in place again

   4.2 Mid point as PLR
   4.2.1 Link Protection (Nhop)
   Objective
     To benchmark the MPLS failover time due to a link failure event experienced by
     mid point PLR. 

   Test Setup
     Use the topologies in Figures 9 through 12.  The DUT is the midpoint PLR.
     The DUT will have one interface connected to the head end
     The DUT will also have 2 other interfaces connected to other test device(s) based 
       on the topology 
     
   Test Configuration
     1. Configure x no of primary and the backup for each of the link protection cases
        as shown in Figures 9 through 12
     2. Provision the tailend to send implicit null label for the primary 
        and backup tunnel
     3. Advertise prefixes (as per FRR Scalability table describe in section 2) by the 
        tail end. 
        
   Procedure 
     1. Configure x no of primary tunel as per FRR Scalability table primary one headend
     2. Configure backup tunnel on DUT
     3. Verify Primary and Backup tunnels are up and that Primary lsps are protected 
     4. Verify Fast reroute protection
     5. Configure 3 traffic stream using any traffic generator. 3 traffic streams should be 
        configure to send traffic to 1st , Mid and Last LSPs from fast reroute database, 
        Make sure do not configure traffic stream in round robin fashion. we need 3 separate 
        stream send to 1st mid and last prefixes
     6. Start traffic as specified in section "Traffic Generation" 
     7. Shutdown Protected interface from remote side (or any other failure trigger) 
     8. Verify Primary tunnel gets mapped to backup tunnels   
     7. Stop traffic stream and measure the traffic loss.
     8. Measure failover time . (Numbers of packet drop/rate per second * 1000)
     9. Start traffic stream again to verify re-optimization when protected interface 
        comes up. Traffic loss should be 0 due to make before break or reversion
     10.Enable protected interface that was shut
     11.Verify headend signals new LSP and Protection should be in place again

   4.2.2 Node Protection (NNhop)
   Objective
     To benchmark the MPLS failover time due to a failure of node downstream to
     midpoint PLR. 

   Test Setup
     Use the topologies in Figures 13 through 16.  The DUT is the midpoint PLR.
     The DUT will have one interface connected to the head end
     The DUT will also have 2 other interfaces connected to other test device(s) based 
       on the topology 
     
   Test Configuration
     1. Configure x no of primary and the backup for each of the link protection cases
        as shown in Figures 13 through 16.
     2. Provision the tailend to send implicit null label for the primary 
        and backup tunnel
     3. Advertise prefixes (as per FRR Scalability table describe in section 2) by the 
        tail end. 
        
   Procedure 
     
     1. Configure x no of primary tunel as per FRR Scalability table primary one headend
     2. Configure backup tunnel on DUT
     3. Verify Primary and Backup tunnels are up and that Primary lsps are protected 
     4. Verify Fast reroute protection
     5. Configure 3 traffic stream using any traffic generator. 3 traffic streams should be 
        configure to send traffic to 1st , Mid and Last LSPs from fast reroute database, 
        Make sure do not configure traffic stream in round robin fashion. we need 3 separate 
        stream send to 1st mid and last prefixes
     6. Start traffic as specified in section "Traffic Generation" 
     7. Reboot or turn off power of the node downstream to the PLR/Ingress (or any other 
        failure trigger)
     8. Verify Primary tunnel gets mapped to backup tunnels   
     7. Stop traffic stream and measure the traffic loss.
     8. Measure failover time . (Numbers of packet drop/rate per second * 1000)
     9. Start traffic stream again to verify re-optimization when protected interface 
        comes up. Traffic loss should be 0 due to make before break or reversion
     10.Enable protected interface that was shut
     11.Verify headend signals new LSP and Protection should be in place again


  5. Security Considerations
       Documents of this type do not directly effect the security of
       the Internet or of corporate networks as long as benchmarking
       is not performed on devices or systems connected to operating
       networks.

  6. Acknowledgements
       Thanks to Amrit Hanspal for all his encouragement. 


_______________________________________________
bmwg mailing list
bmwg at ietf.org
https://www1.ietf.org/mailman/listinfo/bmwg